An EU for AI Deployment in Transport and Mobility

Artificial Intelligence systems have been slowly integrated in transport and mobility throughout the past decades. Today, this process has accelerated drastically with AI pervading these sectors. There are clear benefits for the widespread use of such technology. Still, we must be cautious regarding possible negative impacts technological advancement can have on the EU’s single market, all while eliminating pre-existing market conditions that could stifle innovation.

For these reasons, the European Union must take the lead, focusing on policy which safeguards the respect of EU rules, standards and rights, whilst also promoting competitiveness and ensuring EU transport and mobility enterprises reap the benefits of an AI world.

Undoubtedly, Artificial Intelligence can benefit transport and mobility, in particular when it comes to EU objectives. On road safety, increased automation, remote car checks, driver assistance technologies running on AI can help reduce the human factor which is responsible for 90% of fatal crashes, assisting the EU in achieving its zero road deaths objective by 2050 (Vision Zero).

AI can also contribute to the Green Deal. The integration of AI software in shipping can lead to predict the most fuel-efficient route, helping ships adapt speed and avoid unnecessary emissions. From a mobility perspective, the sharing economy can help reduce emissions through the use of unified platforms (MaaS) promoting multimodality, public transport, and ride sharing.

A less mentioned benefit refers to societal gains of AI for transport. A US study on self-driving cars found that by helping mitigate obstacles to mobility for individuals with disabilities, automation of vehicles can enable employment opportunities for approximately 2 million people and save 19 billion US dollars in healthcare expenditures from missed medical appointments[1].

Lastly, overall, AI will help boost efficiency for transport and logistics by improving internal processes, the digitalization of enterprises, and the overall operations. Through traffic management and better route planning, it can help eliminate congestion, avoiding costs of Euro 100 billion per year. Efficiency gains could lead to a 10% increase of GDP for the sector on average by 2030, greatly benefitting EU businesses[2].

The benefits for the sector are evident. However, many challenges lie ahead. Among the obstacles to the full deployment of AI in transport is lack of harmonized AI rules among Member States. This translates into a slower adoption of AI with negative costs on innovation. Diverging national approaches disproportionally affect SMEs, as they do not possess the means nor finances to spread additional costs of operating in different Member States the same way larger foreign companies can.

The EU’s digital sovereignty ambitions can only be achieved if we remove barriers that impede our businesses in developing AI-driven products. EU policy action for transport must therefore address fragmentation focusing on harmonizing existing policy gaps on liability, safety, security (including cyber), algorithmic transparency and data privacy. Doing so would boost societal acceptance of AI and guarantee legal clarity for EU businesses to invest in the technology.

Let us take liability as an example. How can we expect a boost in AI innovative products’ development and uptake by consumers, if there are no clear rules on who will be held liable in case of an accident? What about cross border uses of a product?

The EU is already taking action on two fronts. First, the Commission is reviewing and updating the legal and policy framework for AI. Secondly, work is being conducted to ensure specific AI enablers are in motion such as investments in technology, data availability and infrastructure. Policy can only do as much. There will be no vast scale AI deployment without massive investments in high performance computing, cloud services, 5G, data platforms and quality data collection. For this reason, the EU has developed programs and joint ventures for transport and mobility such as ERTMS for rail, the SESAR joint undertaking that supports research projects in AI, and the C-Roads platform project for harmonizing Cooperative Intelligent Transport System (C-ITS) deployment in Europe. Still, as per any industrial revolution – we are experiencing the fourth – more is needed.

From a legislative standpoint, the European Commission is working on addressing the gaps in the legal and regulatory framework for AI with three policy actions. The proposed Artificial Intelligence Act to ensure excellence and trust in AI, the upcoming directive on Civil Liability, and various updates to sectoral legislation.

All such measures, by nature, have a horizontal take on the issue. To be clear, we need harmonized legislation to remove barriers, ensure trust, and provide legal clarity. The question therefore must be posed; how can we ensure that such legislation tailors for the specific needs of transport and mobility supporting EU businesses?

First, since transport and mobility are highly regulated, we must avoid overlap between the new AI framework and pre-existing sectoral legislation. For example, clarity is needed on which approval procedures are required for a product embedded with AI.

Second, it is important to avoid unnecessary burden caused by an unjust high-risk classification of an AI system. In the transport sector, some AI applications have been in place for decades. Rule-based AI systems for example are at the basis of automation. Having been in place for years, if proven not of high risks to humans, we must avoid subjecting certain types of AI to new obligations.

Third, we need to ensure that unavoidable burdens do not stifle the competitiveness of EU businesses. We need our champions in transport and mobility, including for the sharing economy. As noted prior, innovation and investments are required if the EU wants to reap the benefits of AI. Legal provisions must be in place, such as regulatory sandboxes and assistance to SMEs, something foreseen by the AI Act, to allow new business models to grow and help the sectors enter the future.

Lastly, if we want to see an uptake in AI use in transport and mobility, we need transparency on how algorithms work and on how consumer data privacy is ensured. Only this will help the consumer attain peace of mind.

The benefits of AI for transport and mobility cannot be questioned; its role in helping us achieve our societal and EU goals neither. Still, if we want to see those benefits materialize effectively, we need coordinated and strengthened EU action for both policy and investments. We need clearer rules on liability, safety of products, security, transparency and data privacy to ensure trust by the public and security of investments for businesses. This is the way Europe can become a global player in AI transport and mobility applications.

[1] https://rudermanfoundation.org/wp-content/uploads/2017/08/Self-Driving-Cars-The-Impact-on-People-with-Disabilities_FINAL.pdf

[2] https://www.europarl.europa.eu/thinktank/en/document/EPRS_STU(2021)654212